|
Title: TRIP GENERATION MODELING USING CHAID, A CRITERION-BASED SEGMENTATION MODELING TOOL
Accession Number: 00757513
Record Type: Component
Record URL: Availability: Transportation Research Board Business Office 500 Fifth Street, NW Find a library where document is available Abstract: Conventional trip generation models are identified, as are the difficulties of model application typical of segmentation problems: identification and categorization of explanatory variables and of the interactions among them. The use of CHAID (Chi-Squared Automatic Interaction Detection), a criterion-based segmentation modeling tool, is explored to analyze household trip generation rates. CHAID models are presented in the form of a tree, each final node representing a group of homogenous households concerning daily trip making. An application to data from an origin-destination survey for Sao Paulo produced interesting results, in agreement with theoretical expectations and amenable to interpretation based on the likely activity-travel patterns of each group of households generated by the technique. CHAID can be used as an exploratory technique for aiding model development or as a model by itself. The use of CHAID results as a trip generation model was verified through an evaluation of its predictive capability in a cross comparison of two subsamples and through a comparison of observed versus predicted trips at a zone level; the segmentation of households produced by the technique provided good estimates of trip rates and zone totals. The application of a modeling approach requiring a highly disaggregate projection of the population may become possible considering the advances in methods for the generation of synthetic populations. The use of these methods in conjunction with a segmentation model represents an alternative to conventional trip generation models and an opportunity to introduce new population forecasting techniques into transportation planning practice.
Supplemental Notes: This paper appears in Transportation Research Record No. 1645, Forecasting, Travel Behavior, and Network Modeling.
Language: English
Corporate Authors: Transportation Research Board 500 Fifth Street, NW Authors: Strambi, Ovan de Bilt, K-APagination: p. 24-31
Publication Date: 1998
Serial: ISBN: 0309065178
Features: Figures
(3)
; References
(30)
; Tables
(2)
TRT Terms: Uncontrolled Terms: Geographic Terms: Old TRIS Terms: Subject Areas: Planning and Forecasting; Transportation (General); I72: Traffic and Transport Planning
Files: TRIS, TRB, ATRI
Created Date: Dec 11 1998 12:00AM
More Articles from this Serial Issue:
|